is_locked optimization

[Jeremy Allison]

On Fri, Dec 10, 2004 at 04:41:00PM +1100, tridge at samba.org wrote:
> 
> Are you really sure about this? Maybe you could write a windows test
> program that uses byte range locks and see if they are not sent on
> oplocked files.

Pretty sure. It's why we got away with the broken locking implementation
in Samba 2.0.x for so long (the one that mapped directly to POSIX locks
remember). Conflicting locks within a client are handled by the client
as an optimization - we never saw them on the wire once we had oplocks.

> Please determine first if the switch potentially causes data
> corruption by writing a win32 test. If it does cause data corruption
> then we should either not have the switch or document very clearly
> that it can cause data corruption.

Ok - will do.

Jeremy.